When data is ingested it is stored in primary storage. Queries of data will be most efficient if the data is already available on primary storage. To maximize query performance, primary drives should be as fast as possible. The tradeoff is that fast storage is expensive, so primary storage is normally extended by optionally adding secondary storage, and bucket storage. Typically, primary storage will be based on fast technology such as NVMe drives. Secondary storage will be slower but cheaper SSDs or magnetic drives, and bucket storage will typically be cloud-based storage such as AWS S3.

The main configuration parameters for primary storage are DIRECTORY, PRIMARY_STORAGE_MAX_FILL_PERCENTAGE, PRIMARY_STORAGE_PERCENTAGE. These enables you to configure the mount point for primary storage, and control the point at which secondary storage is used.

When LogScale is deployed in a cloud environment, promary storage is typically based on NVMe instance stores. These provide maximum performance , but due to the way they are deployed in the cloud environment, should be considered ephemeral storage. This is explained further in the introduction.